Upload
davin
View
78
Download
0
Embed Size (px)
DESCRIPTION
Raimo P. Hämäläinen Systems Analysis Laboratory Aalto University, School of Science Co-authors: Jukka Luoma and Esa Saarinen. On the NEED for behavioral operations research . Behavioral Operations Research. The study of behavioral aspects related to the use of - PowerPoint PPT Presentation
Citation preview
Raimo P. Hämäläinen Systems Analysis Laboratory
Aalto University, School of Science
Co-authors: Jukka Luoma and Esa Saarinen
ON THE NEED FOR BEHAVIORAL OPERATIONS
RESEARCH
Behavioral Operations Research
The study of behavioral aspects related to the use of
operations research methods in modeling, problem solving and
decision support
Behavioral research
How people behave in different settings?What are the consequences of humans being
involved?Research methods: experimental and qualitative
What is the human impact on the OR process?
“Scientific methods to improve the effectiveness of operations and systems to make better decisions”
Scientific methods:Modeling, data analysis, optimization etc.
Operations Research The Science of Better
The pioneers West Churchman and Russel Achoff:
OR is not mathematics only
What is essential in our profession?
Goal to help people in problem solving
but
Have we omitted the people,the problem owners and the OR experts,
from the analysis?
Methods and problem solving
Theory and algorithms are free of behavioral effects
but
as soon as we use them in real life problem solving behavioral effect will be present.
Model validity discussed a lot in early ORThere exists one ideal model and a good OR
specialist needs to find it.
Hidden assumption:A valid model automatically produces a valid
process and bias free objective results
Model validity The lure of objectivity
Best practices in OR Acknowledgement of
subjectivityFocus on the OR process
Based on successful of case studies First steps towards behavioral OR
So far, no behavioral researchHow do the best practices compare against
each other? Can different processes lead to different outcomes? What are the
benefits to the client?
Soft OR and Systems Thinking• Criticized OR for being too narrowly concerned
with mathematical models only• New qualitative methods for framing and
structuring• Attention to the sociology and philosophy of
modeling • Has remained mainly methodology and tool
focused with limited behavioral research
Some areas of OR have a tradition in behavioral studies
Decision and Risk Analysis• Subjectivity is explicitly taken into account• Value and utility functions to describe
preferences• Risk attitudes seeking/averse• Multicriteria evaluation of alternatives with
subjective weighting• Research on biases and
risk perceptions
Operations Management
• Studies how people act in complex decision settings
• Judgemental forecasting• Behavioural operations conference series started
in 2006• The Bullwhip effect in Supply chains - Beer game
Factory Distributor Wholesaler RetailerDelayDelayDelayDelay
Interest in behavioral issues emerges when the basic theoretical core of a
discipline has matured
Behavioral finance and economics
• What is the actual behavior of agents in economic decision making?
• How do people make personal investment decisions?
• Active research area acknowledged also by theoretical economists
• Nobel price 2002 in economics to Vernon Smith together with Daniel Kahneman
Embracing the behavioral perspective in economics
helps:“in generating theoretical insights, making better predictions, and suggesting better
policy” (Colin Camerer et al., 2004)
If this is true for economics it surely applies to OR as well
Judgement and Decision making
From: Kahneman and Tversky
• Decision theory is not enough to explain human choices
• Axioms of rationality not followed• Bounded rationality (Herbert Simon)• Prospect theory: gains and
losses seen differently (Daniel Kahneman and Amos Tversky)
• Cognitive biases• Heuristics (Gerd Gigerenzer)
From behavioral to neural• Emotions are needed in decision making• Somatic marker hypothesis (Antonio Damasio)• Brain imaging research on decision making –
neuroeconomics• How do we evaluate risks - What brain areas are
activated in risk decisions
Experimental Game Theory• How do people interact?• Ultimatum game
• The receiver should accept 1 €, 50% reject offers 20 €
• Strong tendency towards co-operative behaviour• Typically fair offers near 50 euros• Research on reciprocity and fairness• Practical implications on auctions?
Offer x €
100-x € if accept x € 0 € if reject 0 €
Split 100 €
OR is a mature discipline
We are ready to start the behavioral era!
It is natural to pay attention to howhuman behavior moderates the OR
process
OR process creates a system • Formed by the interaction of the client and the
OR analyst – usually a team• The client and the analyst are subject to
behavioral effects• The OR analyst needs to observe and understand
this system to improve its performance• A key to good practice• Use Systems Intelligence i.e. your ability to
successfully and engage with systems (Saarinen and Hämäläinen, 2004)
Social group processes in OR facilitation
• Groupthink – overconfidence (Irving Janis)• Strategic behavior by analyst
and stakeholders • Hidden agendas in modeling:
omission of factors and adverse selection of data
• Gender and cultural effects• Facilitator styles, personality etc.
This is the right model
Yes
Yes
YesYes
YesYes
Problem solving processes• What is the main intended result - learning or
optimizing?• What are the criteria used -optimizing or
satisficing?• How to facilitate when rationality cannot be
enforced?• Human behavior can seem irrational –
intransitive preferences, bounded rationality and path dependence
Research challenge
Comparative experimental research on problem solving and structuring is very difficult
Real problems can seldom be approached repeatedly
with the real decision makersExperiments with students a good first step
OR models of people behavior• People in the loop models – pilots, operators etc.• People behavior in service systems: queuing and
waiting for service • Crowd behavior in emergency situations –
Evacuation in fires, festivals
(From: Ehtamo et al)
OR models of people behavior• People in the loop models – pilots, operators etc.• People behavior in service systems: queuing and
waiting for service • Crowd behavior in emergency situations –
Evacuation in fires, festivals
(From: Ehtamo et al)
We are subject to cognitive biases
• Appeal to Authority: we tend to thoughtlessly obey those (modeling traditions) we regard as being in positions of authority
• Beauty Effect: we attribute qualities to people (models) based on their appearance
• Cognitive Dissonance: the effect of simultaneously trying to believe in two incompatible things (model/real world) at the same time
• Commitment Bias: once we are publicly committed ourselves to a position (model) we find it difficult to retreat
• Confirmation Bias: we interpret evidence to support our prior beliefs (models)
• Fundamental Attribution Error: we attribute success to our own skill (model) and failure to everyone else's skill (rivaling models)
• Inter-group Bias: we evaluate people within our own group (modelling tradition) more favorably than those outside of it
• Loss Aversion: we do stupid things to avoid realizing a loss (acknowledging failure of our modelling)
• Man With A Hammer Syndrome: some people have a single tool (model) and see every problem as a nail
• Overconfidence: we're way too confident in our abilities (models)
• Priming: exposure to some event (modelling approach) changes our response to a later event (problem needing another model)
• Representative Heuristic: we compare the under consideration (modelling approach) to whatever we happen to bring to mind
Behavioral studies in OR aim to find ways to reveal and avoid cognitive
biases in the OR process
Framing• Increasingly important when moving from
optimization to solving people related problems• Behavioral elements are strong• Definition of system boundaries and stakeholders• Stakeholders have different perspectives and
mental models• Creating a common language• A key step in many environmental problems
Model building• Usefulness of simple versus complex models• How to build models to maximize learning• Anchoring effect in selecting model scale and
reference point• Are prospect theory related phenomena relevant
when choosing the sign (increasing/decreasing) of variables
Communication with and about models
• Visual representation of system models are essential in communication
• Effects of graphs and scales used • What is the effect of educational and cultural
backgrounds of the problem owners• What can we learn from statistics?• Is software development based on behavioral
studies?
Effect of Graphical Interfaces and
Example:Simulation
Mathematica System Modeler
Vensim
True
Matlab Simulink
Behavioral research topics in OR
Teaching of OR• Balance between methods and people skills• Should every OR student learn behavioral issues?• How to teach best practices?• Developing facilitation and systems intelligence
skills• Role of software
Ethics and OR
• Ethical OR takes behavioral challenges seriously• OR is used in the most important problems of
mankind – climate models and policies• Unintentional biases in model use• Are we really solving the problem or selling our
model?• How to improve self leadership skills in OR
practice
Non-expert use of OR methods
• Modelling is a tool used in many fields• Easy OR software invites non-experts• What is the result?• What are the typical pitfalls and risks?• Who should supervise the use of OR models?• Is quick learning of the OR process possible?• Collaboration between experts and non-experts
ExampleBehavioral studies in system
dynamics
Understanding dynamics in climate change is important in modern world (John Sterman, MIT)
Why don’t well-educated adults understand
accumulation? A challenge to researchers, educators
and citizensCronin, Gonzalez, Sterman (2009)• Accumulation refers to the growth of a stock
variable when the inflow exceeds the rate of outflow
• Carbon dioxide in the atmosphere, Balance of bank accounts, Milk in the refrigerator etc.
• Experiments with the Department store task with MIT students
People entering and leaving the department store
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300
5
10
15
20
25
30
35
40
Minute
Peop
le /
min
ute
entering leaving
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300
5
10
15
20
25
30
35
40
Minute
Peop
le /
min
ute
entering leaving
During which minute did the most people enter the store?
During which minute did the most people enter the store?
96% correct answers
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300
5
10
15
20
25
30
35
40
Minute
Peop
le /
min
ute
entering leaving
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300
5
10
15
20
25
30
35
40
Minute
Peop
le /
min
ute
entering leaving
During which minute were the most people in the store?
During which minute were the most people in the store?
44% correct
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300
5
10
15
20
25
30
35
40
Minute
Peop
le /
min
ute
entering leaving
During which minute were the fewest people in the
store? 31% correct
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300
5
10
15
20
25
30
35
40
Minute
Peop
le /
min
ute
entering leaving
Wrong CorrectWrong
Easy to adopt a misleading starting frame
• General stock and flow system – try the general procedure and integrate the difference between the inflow and the outflow
• The department store task is a simple special case
• Computation is not required • Observe the fact that the inflow and outflow
curves intersect only once• The correct answer is obvious
Behavioural problems
• False cues which mislead the participants• Questions do not address accumulation directly• Shapes of the curves trigger inappropriate
heuristics• Availability heuristic: maximum, inflow and
outflow stand out• “Cannot be determined,” box primes to think the
task is very difficult
Re-examining the experimentAalto University students in Finland
I. Repetition of MIT procedure • Similar results
II. Revised questionnaire • Smoother curves to reduce the impact of
availability heuristic• Added questions asking about the accumulation
phenomenon directly
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300
5
10
15
20
25
30
35
Minute
Peop
le /
min
ute
Revised smoother curves
leavingentering
Almost all of the participants were able to understand
accumulation “During which minute were the most people in the store?” (88-90% correct – originally 44%)“During which minute were there the fewest people in the store?” (72 - 76% correct – originally 31%)
People’s poor performance in the department store task does not reflect the existence of a new cognitive bias as suggested by Cronin et al.
Lesson learnt
Even the simple accumulation phenomenon can be misunderstood in the presence of distacting
triggers of biases
Extreme care needed when communicating about systems and models
SummaryBehavioral aspects influence the OR process
Framing, biases, communication, learning, group processes
The practice of OR can be improved by behavioral research
Using the term Behavioral OR will stimulate research
Behavioral OR needs to be recognized as an integral part of OR
Behavioral OR could take a leading role in advancing the responsible use of models in policy
issuesA mature field like OR becomes stronger with
behavioral research
Developing practitioner skills with a behavioral lens will keep OR alive and interesting
for our customers and the society at large
Thank you!
References and linksPresentation based on paper: R.P. Hämäläinen, J. Luoma and E. Saarinen: On the Importance of Behavioral Operational Research: The Case of Understanding and Communicating about Dynamic System, European Journal of Operational Research 2013, Vol. 228, Issue 3, pp. 623-634.
References:R.L. Ackoff: Some unsolved problems in problem solving. Operational Research Quarterly, 13:1-11, 1962C.F. Camerer and G. Loewenstein: Behavioral economics: Past, Present Future in: C.F. Camerer, G. Loewenstein, M. Rabin: Advances in Behavioral Economics, Princeton University Press, pp. 3-51. 2004. C.W. Churchman: Operations research as a profession. Management Science, 1970, 17(2), B37-B-53.M.A. Cronin, C. Gonzalez and J.D. Sterman: Why don’t well-educated adults understand accumulation? A challenge to researchers, educators, and citizens. Organizational Behavior and Human Decision Processes, 2009, 108(1), 116-130.A.R. Damasio: Descartes' Error: Emotion, Reason, and the Human Brain, London, Vintage,1994.H. Ehtamo, S. Heliövaara, T. Korhonen and S. Hostikka: Game Theoretic Best-Response Dynamics for Evacuees' Exit Selection Advances in Complex Systems, 2010, 13(1), 113-134.G. Gigerenzer, P.M. Todd and the ABC Group: Simple heuristics that make us smart, New York: Oxford University Press, 1999.R.P. Hämäläinen and E. Saarinen: Systems intelligence - the way forward? A note on Ackoff’s “Why few organizations adopt systems thinking.” Systems Research and Behavioral Science, 2008, 25(6), 821-825.
I. Janis: Groupthink: Psychological Studies of Policy Decisions and Fiascoes , Wadsworth, USA,1982.I. P. Levin, S.L. Schneider and G.J. Gaeth: All Frames Are Not Created Equal: A Typology and Critical Analysis of Framing Effects,. Organizational Behavior and Human Decision Processes, 1998, 76(2), 149-188. J. Luoma, R.P. Hämäläinen and E. Saarinen: Acting with systems intelligence: integrating complex responsive processes with the systems perspective. Journal of the Operational Research Society, 2010, 62(1), 3-11.E. Saarinen and R.P. Hämäläinen: Systems Intelligence: Connecting Engineering Thinking with Human Sensitivity. Systems Intelligence: Discovering a Hidden Competence in Human Action and Organizational Life, Systems Analysis Laboratory Research Reports. Helsinki University of Technology, 2004.H. Simon: Models of Bounded Rationality, Vol. 1. MIT Press, 502 pp,1997. J.D. Sterman: Modeling Managerial Behavior: Misperceptions of Feedback in a Dynamic Decision Making Experiment. Management Science, 1989, 35(3), 321-339. J.D. Sterman: Economics: Risk Communication on Climate: Mental Models and Mass Balance. Science, 2008, 322(5901), 532-533. D. von Winterfeldt and W. Edwards: Decision analysis and behavioral research (Vol. 1),1986, Cambridge University Press.
Systems Intelligence Research Groupwww.systemsintelligence.tkk.fi/